The splicing together of exons during the transcription of most eukaryotic pre-mRNA molecules is a fundamental step in the transfer of information from DNA to protein. However, how the splice sites are recognized to initiate this fast and accurate process in not at all clear, since false sequences that resemble real sites outnumber the latter by 1 or 2 orders of magnitude yet are not used. We are searching for the additional information that specifies splice site recognition during pre-mRNA splicing in mammalian cells. In previous work we have used computational analysis to discover that: 1) that exon flanks 50 nt beyond the splice site consensuses can greatly influence splicing and contain sequences that stand out from the background; and 2) there are a large number of specific 8 mers that can act either as exonic splicing enhancers or splicing silencers. We will characterize both of these types of sequences, defining their base requirements, extent, position dependence, and specificity. With regard to the last, we will develop a novel exhaustive survey method that uses massively parallel solid state sequencing to identify every possible 8-mer that can be functional in a particular context, and we will compare this profile in different intronic, exonic and cellular contexts for constitutive and alternatively spliced exons. The flanking sequences fall into 2 distinct classes according to their G+C contents; we will test the idea that the rules governing exon definition differ for the two classes. To simplify the discovery of rules that govern exon definition, we will create designer exons made up of synthetic 8-mer modules we have identified as enhancers, silencers or neutral sequences. The numbers and spacing of these modules will reveal relationships that are hidden in the complexity of natural sequences. Finally, we will continue to use computation, including new machine learning algorithms, to better define intronic elements and to discover interactions between two or more features of a local sequence that define a splice site. We will also develop a Web-based exon-finding program that integrates all the information we have accumulated. Pre-mRNA splicing goes awry in a large proportion or human genetic disease cases. An understanding of the rules that govern the recognition of splice sites should help in the design of therapeutic intervention strategies to reverse such ill effects. [unreadable] [unreadable] [unreadable] [unreadable] [unreadable] [unreadable] [unreadable] [unreadable]